Transcript Powerpoint Slides - Suraj @ LUMS

MPLS Protection Routing:
A Tutorial
Zartash Afzal Uzmi
First slide…
Questions?
Ask when you have them!
Jan 13, 2006
Lahore University of Management Sciences
2
Outline


Background
 Network Services and QoS
 Architectural Requirements
 IP and MPLS
Introduction to protection and restoration routing

Terminology
Local Protection: Types of Backup Paths Fault Models
 Backup Bandwidth Sharing
 Activation sets
Protection routing framework
 Components
 Typical example
 Evaluation and Experimentation


Jan 13, 2006
Lahore University of Management Sciences
3
Outline



Background
 Network Services and QoS
 Architectural Requirements
 IP and MPLS
Introduction to protection and restoration routing
 Terminology
 Local Protection: Types of Backup Paths Fault Models
 Backup Bandwidth Sharing
 Activation sets
Protection routing framework
 Components
 Typical example
 Evaluation and Experimentation
Jan 13, 2006
Lahore University of Management Sciences
4
Network Traffic and Services

Network Traffic today



New and interactive applications are emerging





Not what it was 10 years ago
Multimedia intensive
Internet telephony
Videoconferencing
Streaming media (voice and video)
Remote collaboration (e.g., remote desktop)
Many new applications are real-time

More and more users of these applications
Burstiness behavior has changed over the years!
Jan 13, 2006
Lahore University of Management Sciences
5
Current Network Architecture

Internet is popular because


Internet is inexpensive because


It uses resource sharing
 by means of statistical multiplexing
Current Internet architecture



It is inexpensive
Uses packet switches with buffers
Required buffer size is primarily determined by a
random traffic pattern
Buffer size optimization


Too low  High drop rate
Too high  High delay
Jan 13, 2006
Lahore University of Management Sciences
6
Architectural Requirements

Emerging applications



Under normal conditions


We are worried about the buffers used in two-way
interactive applications
When resources fail


Two-way interactive communications
One-way streaming media type applications
We are also worried about the one-way applications
Current Internet architecture is not suitable for
new and emerging applications

New architectures are being researched
Jan 13, 2006
Lahore University of Management Sciences
7
Architectural Requirements

New network architectures



Experience with networks



All circuit-switched?
Mix of packet-switch and “circuit-switch-like”
Bigger buffers are required when there is more
randomness and more aggregation
Should use circuits at places where we see more
aggregation
Example: 100x100 project


Edge network is packet-switched
Core network is virtual-circuits
Jan 13, 2006
Lahore University of Management Sciences
8
IP versus MPLS


In IP Routing, each router makes its own routing
and forwarding decisions
In MPLS:




source router makes the routing decision
Intermediate routers make forwarding decisions
A path is computed and a “virtual circuit” is established
from ingress router to egress router
An MPLS path or virtual circuit from source to
destination is called an LSP (label switched path)
Jan 13, 2006
Lahore University of Management Sciences
9
Outline


Background
 Network Services and QoS
 Architectural Requirements
 IP and MPLS
Introduction to protection and restoration routing

Terminology
Local Protection: Types of Backup Paths Fault Models
 Backup Bandwidth Sharing
 Activation sets
Protection routing framework
 Components
 Typical example
 Evaluation and Experimentation


Jan 13, 2006
Lahore University of Management Sciences
10
Protection and Restoration

Restoration



Protection



Pre-determined recovery – backup paths “in advance”
Primary and backup are provisioned at the same time
IP supports restoration


On-demand recovery – no preset backup paths
Example: existing recovery in IP networks
Because it is datagram service
MPLS supports restoration as well as protection

Because it is virtual-circuit service
Jan 13, 2006
Lahore University of Management Sciences
11
Restoration in IP network

In traditional IP, what happens when a link
or node fails?




Failure information needs to be disseminated in
the network
During this time, packets may go in loops
Restoration latency is in the order of seconds
We look for protection possibilities in an
MPLS network, but…

First we need to look at the QoS requirements
Jan 13, 2006
Lahore University of Management Sciences
12
QoS Requirements

Bandwidth Guaranteed Primary Paths

Bandwidth Guaranteed Backup Paths


BW remains provisioned in case of network failure
Minimal “Protection or Restoration Latency”

Protection/Restoration latency is the time that elapses
between:
 “the occurrence of a failure”, and
 “the diversion of network traffic on a new path”
Restoration is generally SLOWER than protection
Jan 13, 2006
Lahore University of Management Sciences
13
Protection in MPLS



First we define Protection level
Path protection
 Also called end-to-end protection
 For each primary LSP, a node-disjoint backup LSP is set up
 Upon failure, ingress node diverts traffic on the backup path
Local Protection
 Upon failure, node immediately upstream the failed element
diverts the traffic on a “local” backup path
Path Protection  More Latency
Local Protection  Less Latency
Jan 13, 2006
Lahore University of Management Sciences
14
Protection in MPLS
Path Protection
S
1
2
3
D
This type of “path Protection”
still takes 100s of ms.
Primary Path
Backup Path
Jan 13, 2006
We may explore “Local Protection” to
quickly switch onto backup paths!
Lahore University of Management Sciences
15
Local Protection: Fault Models
Link
Protection
Node
Protection
Element
Protection
Jan 13, 2006
A
B
C
D
A
B
C
D
A
B
C
Lahore University of Management Sciences
D
16
Protection Modes

1+1 protection



1:1 protection


A backup path protects a single LSP (or a portion of a
single LSP)
N:1 protection




Flow sent on two separate disjoint paths
Receiver responsible for choosing one of the two
A backup path protects one link or one node or both
Overlapping portions of many LSPs are protected by a
single backup path
Applicable for local protection only
N:M protection (M<N)
Jan 13, 2006
Lahore University of Management Sciences
17
nhop and nnhop paths
LOCAL PROTECTION
nnhop
A
B
D
C
E
nhop
PLR: Point of Local Repair
Primary Path
Backup Path
Jan 13, 2006
All links and all nodes are protected!
Lahore University of Management Sciences
18
Opportunity cost of backup paths

Local Protection requires that backup paths are setup in
advance

Upon failure, traffic is promptly switched onto preset
backup paths

Bandwidth must be reserved for all backup paths


This results in a reduction in the number of Primary LSPs
that can otherwise be placed on the network
Can we reduce the amount of “backup bandwidth” but
still provide guaranteed backups?
Jan 13, 2006
Lahore University of Management Sciences
19
BW Sharing in backup Paths

Example:
Sharing
LSP1
A
BW: X
B
X
X
E
LSP2
G
F
X+Y
Y
C
X
max(X, Y)
BW: Y
Y
D
Primary Path
Backup Path
Jan 13, 2006
Lahore University of Management Sciences
20
Activation Sets
A
A
E
B
C
C
D
D
Activation set for node B
Jan 13, 2006
E
B
Activation set for link (A,B)
Lahore University of Management Sciences
21
Outline



Background
 Network Services and QoS
 Architectural Requirements
 IP and MPLS
Introduction to protection and restoration routing
 Terminology
 Local Protection: Types of Backup Paths Fault Models
 Backup Bandwidth Sharing
 Activation sets
Protection routing framework
 Components
 Typical example
 Evaluation and Experimentation
Jan 13, 2006
Lahore University of Management Sciences
22
Protection Routing Frameworks



We look to answer the following questions?
 Who computes the primary path?
 What is the fault model (link, node, or element protection)?
 Where do the backup paths originate?
 Who computes the backup path?
 At what point do the backup paths merge back with the
primary path
 What information is stored locally in the nodes/routers
 What information is propagated through routing protocols
 What if a primary path can not be fully protected
The goal is almost always to maximize bandwidth sharing
Performance criteria is almost always the maximum number
of primary LSPs that can be placed on the network
Jan 13, 2006
Lahore University of Management Sciences
23
Evaluation & Experimentation

Traffic Generation



Rejected Requests Experiments




Use existing or emerging traffic models
Consider call holding times and multi-service traffic
Generate a set of LSP requests
Measure the number of rejected requests
Simulate on various topologies
Network Loading Experiments



Set link capacities to infinity
Measure the total bandwidth required to service a given
set of LSP requests
Simulate on various topologies
Jan 13, 2006
Lahore University of Management Sciences
24
Recent Trends


Preemption of lower class traffic
Multilayer recovery



We can “almost” deal with recovery at a single protocol
layer
What if we intend to provide recovery at multiple
protocol layers?
For multilayer recovery, we need to consider
these additional issues:





Interworking of layers
Local information stored at each node of each layer
Recovery provided by each individual layer
Signaling mechanism from one layer to another
Effects on bandwidth sharing (if sharing is used)
Jan 13, 2006
Lahore University of Management Sciences
25
We are not done, yet…
Questions & Answers
Jan 13, 2006
Lahore University of Management Sciences
26
Extra Stuff!
Example: A Protection Routing
Architecture
Jan 13, 2006
Lahore University of Management Sciences
27
Extent of BW Sharing: oAIS
More Information propagated  More potential for BW sharing

Aggregate Information Scenario (AIS)




Fij: Bandwidth reserved on link (i, j) for all primary LSPs
Gij: Bandwidth reserved on link (i, j) for all backup LSPs
Rij: Bandwidth remaining on link (i, j)
Optimized AIS (oAIS) – (Hij instead of Fij)


Hij:
Maximum bandwidth reserved on any one link by all
backup paths spanning link (i, j)
Also propagate Gij and Rij
Jan 13, 2006
Lahore University of Management Sciences
28
oAIS versus AIS: Example
LSP Request-1 (src, dst, bw) = (A, C, 4)
D
GAF=4
E
F
FAB=4
A
HAB=4
B
C
G
Jan 13, 2006
Lahore University of Management Sciences
29
oAIS Example
LSP Request-2 (src, dst, bw) = (A, C, 5)
D
GAF=4
E
F
FAB=4
=9
A
B
C
HAB=4
=5
GAG=5
G
Jan 13, 2006
Lahore University of Management Sciences
30
oAIS Example
LSP Request-3 (src, dst, bw) = (D, E, 7)
FDE=7
D
GAF=7
=4
E
F
FAB=9
A
B
C
HAB=5
GAG=5
G
Jan 13, 2006
Lahore University of Management Sciences
31
oAIS Example
LSP Request-4 (src, dst, bw) = (A, C, 6)
Need to Evaluate cost of all possible backup paths?
How much BW is shareable on (A, F)?
D
AIS:
Shareable = max(0, GAF - FAB)
= GAF - min(GAF, FAB) = 0
Additional resv = 6
oAIS: (HAB ≤ FAB)
Shareable = GAF - min(GAF, HAB) = 2
Additional resv = 6 - 2 = 4
CIS: (link (A,B) knows BWred)
Shareable = GAF - BWred = 7 - 4 = 3
Additional resv = 6 - 3 = 3
Jan 13, 2006
GAF=7
FDE=7
E
F
FAB=9
A
B
C
HAB=5
GAG=5
Lahore University of Management Sciences
G
32
Single Link Protection: Network 1
Jan 13, 2006
Lahore University of Management Sciences
33
Single Link Protection: Network 1
Jan 13, 2006
Lahore University of Management Sciences
34
Single Link Protection: Network 2
Jan 13, 2006
Lahore University of Management Sciences
35
Single Link Protection: Network 2
Jan 13, 2006
Lahore University of Management Sciences
36
Single Node Protection: Network 1
Jan 13, 2006
Lahore University of Management Sciences
37
Single Element Protection: Network 1
Jan 13, 2006
Lahore University of Management Sciences
38
More Extra Stuff!
Bandwidth Sharing Model for
oAIS
Jan 13, 2006
Lahore University of Management Sciences
39
A Bandwidth Sharing Model
(Simplified for the Link Protection Fault Model)
Recall the definition of nhop paths
Link
Protection
A
B
C
D
Primary Path
Backup Path
Jan 13, 2006
All links are protected!
Lahore University of Management Sciences
40
Bandwidth Sharing Model

Previous:



Aij:= Set of all primaries traversing through (i, j)
Buv:= Set of all backups traversing through (u, v)
New definition (specialized for link protection
case):




Aij:= Set of all primaries traversing through (i, j)
Buv:= Set of all nhop paths traversing through (u, v)
µij:= Set of all nhop paths that span (i, j)
ijuv:= Buv ∩ µij (set of paths falling on (u,v) if (i,j) fails)
Jan 13, 2006
Lahore University of Management Sciences
41
Bandwidth Sharing Model
u
3
i
OLD MODEL:
Aij = {R, B}
Buv = {R, B, …}
Aij ∩ Buv= {R, B}
|| Aij ∩ Buv || = 2+7 = 9
Un-shareable = 9
Shareable = 10 - 9 = 1
Jan 13, 2006
v
j
RED=7
BLU=2
GRN=3 (New Request)
Guv = 10
k
NEW MODEL:
Aij = {R, B}
Buv = {nhijr, nhijb, …}
(nhops through (u, v))
µij = {nhijr, nhijb, …}
(nhops spanning (i, j))
ijuv = µij ∩ Buv= {nhijr, nhijb}
|| ijuv || = 2 + 7 = 9
(Un-shareable)
Shareable = Guv - || ijuv || = 10 - 9 = 1
Lahore University of Management Sciences
42
Bandwidth Sharing Model
u
3
i
OLD MODEL:
Aij = {R, B}
Buv = {R, B, …}
Aij ∩ Buv= {R, B}
|| Aij ∩ Buv || = 2+7 = 9
Un-shareable = 9
Shareable = 10 - 9 = 1
Jan 13, 2006
v
j
RED=7
BLU=2
GRN=3 (New Request)
Guv = 10
k
NEW MODEL:
Aij = {R, B}
Buv = {nhijr, nhjkb, …}
(nhops through (u, v))
µij = {nhijr, nhijb, …}
(nhops spanning (i, j))
ijuv = µij ∩ Buv= {nhijr}
|| ijuv || = 7
(Un-shareable)
Shareable = Guv - || ijuv || = 10 - 7 = 3
Lahore University of Management Sciences
43
Last slide…
Thank you!
Questions?
Jan 13, 2006
Lahore University of Management Sciences
44